Generalized Global Symmetries
Nathan Seiberg
Institute for Advanced Study
Ofer Aharony, NS, Yuji Tachikawa, arXiv:1305.0318
Sergei Gukov, Anton Kapustin, arXiv:1307.4793
Anton Kapustin, Ryan Thorngren, arXiv:1308.2926, arXiv:1309.4721
Anton Kapustin, NS, arXiv:1401.0740
Davide Gaiotto, Anton Kapustin, Shlomo Razamat, NS, Brian Willett,
unpublished
Ordinary global symmetries
• Generated by operators associated with co-dimension one
manifolds 𝑀
𝑈𝑔 𝑀
𝑔 ∈ 𝐺 a group element
• The correlation functions of 𝑈𝑔 𝑀 are topological!
• Group multiplication 𝑈𝑔1 𝑀 𝑈𝑔2 𝑀 = 𝑈𝑔1 𝑔2 𝑀
• Local operators 𝑂 𝑝 are in representations of 𝐺
𝑗
𝑈𝑔 𝑀 𝑂𝑖 𝑝 = 𝑅𝑖 𝑔 𝑂𝑗 𝑝
where 𝑀 surrounds 𝑝 (Ward identity)
• If the symmetry is continuous,
𝑈𝑔 𝑀 = 𝑒 𝑖 ∫ 𝑗 𝑔
𝑗(𝑔) is a closed form current (its dual is a conserved current).
𝑞-form global symmetries
• Generated by operators associated with co-dimension
𝑞 + 1 manifolds 𝑀 (ordinary global symmetry has 𝑞 = 0)
𝑈𝑔 𝑀
𝑔 ∈ 𝐺 a group element
• The correlation functions of 𝑈𝑔 𝑀 are topological!
• Group multiplication 𝑈𝑔1 𝑀 𝑈𝑔2 𝑀 = 𝑈𝑔1 𝑔2 𝑀 .
Because of the high co-dimension the order does not matter
and 𝐺 is Abelian.
• The charged operators V 𝐿 are on dimension 𝑞 manifolds 𝐿.
Representations of 𝐺 – Ward identity
𝑈𝑔 𝑀 𝑉 𝐿 = 𝑅 𝑔 𝑉 𝐿
where 𝑀 surrounds 𝐿 and 𝑅(𝑔) is a phase.
𝑞-form global symmetries
If the symmetry is continuous,
𝑈𝑔 𝑀 = 𝑒 𝑖 ∫ 𝑗
𝑔
𝑗(𝑔) is a closed form current (its dual is a conserved current).
Compactifying on a circle, a 𝑞-form symmetry leads to a 𝑞-form
symmetry and a 𝑞 − 1-form symmetry in the lower dimensional
theory.
• For example, compactifying a one-form symmetry leads to an
ordinary symmetry in the lower dimensional theory.
No need for Lagrangian
• Exists abstractly, also in theories without a Lagrangian
• Useful in dualities
𝑞-form global symmetries
• Charged operators are extended (lines, surfaces)
• Charged objects are extended – branes (strings, domain walls)
– In SUSY BPS bound when the symmetry is continuous
• Selection rules on amplitudes
• Couple to a background classical gauge field (twisted
boundary conditions)
• Gauge the symmetry (sum over these background fields)
• The symmetry could be spontaneously broken
• There can be anomalies and anomaly inflow on walls
Example 1: 4𝑑 𝑈(1) gauge theory
Two global 𝑈 1 one-form symmetries:
• Electric symmetry
– Closed form currents:
2
𝑔2
∗ 𝐹 (measures the electric flux)
– Shifts the gauge field 𝐴 by a flat connection
• Magnetic symmetry
– Closed form currents:
1
𝐹
2𝜋
(measures the magnetic flux)
– Shifts the magnetic gauge field by a flat connection.
Example 1: 4𝑑 𝑈(1) gauge theory
The symmetries are generated by surface operators
𝑈𝑔𝐸 =𝑒 𝑖𝛼 , 𝑔𝑀 =𝑒 𝑖𝜂 𝑀 =
𝑖𝜂
2𝑖𝛼
∫
𝐹
+
2 ∫∗𝐹
2𝜋
𝑔
𝑒
• These are Gukov-Witten surface operators (rescaled 𝛼, 𝜂).
• They measure the electric and the magnetic flux through the
surface 𝑀.
The charged objects are dyonic lines
𝑊𝑛 𝐿 𝐻𝑚 (𝐿)
(𝑊𝑛 𝐿 are Wilson lines and 𝐻𝑚 𝐿 are ‘t Hooft lines)
with global symmetry charges 𝑛 and 𝑚 under the two global
𝑈 1 one-form symmetries.
Example 2: 4𝑑 S𝑈(𝑁) gauge theory
• Electric 𝒁𝑁 one-form symmetry
– The Gukov-Witten operator is associated with a conjugacy
class in 𝑆𝑈(𝑁). When this class is in the center of 𝑆𝑈(𝑁)
the surface operator is topological.
– It shifts the gauge field by a flat 𝒁𝑁 connection.
– It acts on the Wilson lines according to their
representation under the 𝒁𝑁 ∈ 𝑆𝑈(𝑁) center.
• No magnetic one-form symmetry.
– In this theory there are no ‘t Hooft lines – they are not
genuine line operators – they need a surface.
– An open surface operator, whose boundary is an ‘t Hooft
line.
Example 3: 4𝑑 S𝑈(𝑁) gauge theory
with matter in 𝑵
The presence of the charged matter explicitly breaks the electric
one-form 𝒁𝑁 symmetry.
Hence, there is no global one-form symmetry.
Significance of these symmetries
• Consequence: selection rules, e.g. in compact space the vev
of a charged line wrapping a nontrivial cycle vanishes
[Witten].
• Dual theories must have the same global symmetries. (They
often have different gauge symmetries.)
– The one-form symmetries are typically electric on one side
of the duality and magnetic on the other.
– 4𝑑 𝑁 = 1 SUSY dualities respects the global symmetries.
– The 𝑆𝐿(2, 𝒁) orbit of a given 𝑁 = 4 theory must have the
same global symmetry.
Significance of these symmetries
• Twisted sectors by coupling to background gauge fields
– An 𝑆𝑈(𝑁) gauge theory without matter can have twisted
boundary conditions – an 𝑆𝑈(𝑁)/𝒁𝑁 bundle, which is not
an 𝑆𝑈(𝑁) bundle – ‘t Hooft twisted boundary conditions.
• Gauging the symmetry by summing over twisted sectors – like
orbifolds.
– Discrete 𝜃-parameters are analogs of discrete torsion.
• Can characterize phases of gauge theories by whether the
global symmetry is broken or not…
Characterizing phases
• In a confining phase the electric one-form symmetry is
unbroken.
– The confining strings are charged and are classified by the
unbroken symmetry.
• In a Higgs or Coulomb phase the electric one-form symmetry
is broken.
– Renormalizing the perimeter law to zero, the large size
limit of 𝑊 is nonzero – vev “breaks the symmetry.”
– It is unbroken in “Coulomb” phase in 3𝑑 and 2𝑑.
Example 1: 4𝑑 𝑈(1) gauge theory
• There are two global 𝑈 1 one-form symmetries.
• Both are spontaneously broken:
– The photon being their Goldstone boson
0 𝐹𝜇𝜈 𝜖, 𝑝 = (𝜖𝜇 𝑝𝜈 − 𝜖𝜈 𝑝𝜇 )𝑒 𝑖 𝑝𝑥
– Placing the theory on 𝑹3 × 𝑺1 , each one-form global
symmetry leads to an ordinary global symmetry and a oneform symmetry.
– These ordinary symmetries are manifestly spontaneously
broken – the moduli space of vacua is 𝑻2 parameterized by
𝐴4 and the 3𝑑 dual photon.
Example 2: 4𝑑 S𝑈(𝑁) gauge theory
In the standard confining phase the electric 𝒁𝑁 one-form
symmetry is unbroken.
• Charged strings
• Area law in Wilson loops
• When compactified on a circle an ordinary (𝑞 = 0) 𝒁𝑁 , which
is unbroken [Polyakov]
If no confinement, the global 𝒁𝑁 symmetry is broken
• No charged strings
• Perimeter law in Wilson loops
• When compactified on a circle an ordinary (𝑞 = 0) 𝒁𝑁 , which
is broken [Polyakov]
Example 2: 4𝑑 S𝑈(𝑁) gauge theory
Can also have a phase with confinement index 𝑡, where the
global one-form symmetry is spontaneously broken 𝒁𝑁 → 𝒁𝑡 .
• 𝑊 has area law but 𝑊 𝑡 has a perimeter law [Cachazo, NS,
Witten].
• In this case there is a 𝒁𝑁/𝑡 gauge theory at low energies – long
range topological order.
Example 3: 4𝑑 S𝑈(𝑁) gauge theory
with matter in 𝑵
No global one-form symmetry.
Hence we cannot distinguish between Higgs and confinement.
This is usually described as screening the loop [Fradkin, Shenker;
Banks, Rabinovici].
From our perspective, due to lack of symmetry.
Conclusions
• Higher form global symmetries are ubiquitous.
• They help classify
– extended objects (strings, domain walls, etc.)
– extended operators/defects (lines, surfaces, etc.)
• As global symmetries, they must be the same in dual theories.
• They extend Landau characterization of phases based on
order parameters that break global symmetries.
– Rephrase the Wilson/’t Hooft classification in terms of
broken or unbroken one-form global symmetries.
• Anomalies
– ‘t Hooft matching conditions
– Anomaly inflow
– Degrees of freedom on domain walls
Thank you for your attention